Cross-modal Retrieval on MSR-VTT 3 modal
29GapCLIP
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CLIPEvaluation Protocol=LT2026.01 | 29 | 34.2 | 10.3 | — | — | |
| CLIP (LT)Evaluation protocol=Locked-Tuning (LT), Modality configuration=3 modal2026.01 | 24 | — | — | 27 | 30.6 | |
| CLIPEvaluation Protocol=FT2026.01 | 19 | 34.8 | 10.1 | — | — | |
| CLIP (FT)Evaluation protocol=Fine-Tuning (FT), Modality configuration=3 modal2026.01 | 17 | — | — | 37 | 30.3 | |
| NotAGapModality configuration=3 modal2026.01 | 9 | — | — | 7 | 29.9 | |
| ModGapEvaluation Protocol=Ours2026.01 | 7 | 32.8 | 11.8 | — | — | |
| OursModality configuration=3 modal2026.01 | 6 | — | — | 53 | 33.3 |